Write "Random models of Boolean networks"

[daemontus]

See #55 for context.

This should describe various random models of Boolean networks that we are considering.

The basis are N-K networks. However, we can probably just fix K=2 from the beginning, as this is where the critical transition occurs.

However, we need to differentiate between networks with constant in-degree and exponential/Poisson in-degree.

The second aspect is the considered update functions. The "baseline" are completely random functions. However, we would like to also test something more biologically realistic like locally-monotonic functions. Since these are hard to generate randomly, we might want to consider something like threshold old nested-canalising functions.